Anti-shielding of a crack tip by a dislocation is examined at the atomistic level for a simple geometry to test classical singular-crack and recent cohesive-crack models of crack/dislocation interactions. The atomistic model shows that, as an anti-shielding dislocation approaches the crack tip, it causes less anti-shielding than predicted by the singular-crack model. The trend is qualitatively consistent with predictions of a cohesive-crack model, but the atomistic effect is even larger. The cohesive-crack model is consistent with the atomistic results if a reduced cohesive strength of $3.5 GPa is used instead of the actual value of 13 GPa. The difference is shown to be due to the non-linear deformation of material around the crack tip, which cannot be fully represented by a cohesive zone law along the fracture surface. It is then shown that, at the point of fracture, there is a unique traction-displacement cohesive law acting behind the crack tip, independent of the position of the anti-shielding dislocation. The maximum traction of 12.8 GPa and fracture energy of 1.9 J m Γ2 are both in excellent agreement with the values obtained from independent atomistic calculations on this material. Both the shielding and cohesive results have implications for the accurate modeling of fracture processes in metallic materials.